You ask a lot of different questions.
I would say that as a whole, an automated tool that produces 100% accurate results is impossible. Else all the penetration testers in the world would be out of a job.
To give one example, RedHat backports security patches to older versions of software under their long-term support. An automated tool might detect that the server is running a particular kernel version that has a privilege escalation exploit and report it as such. However, due to RedHat's security patches, that particular privilege escalation vulnerability is fixed and no longer exploitable. A penetration tester however, would use that automated tool's report as a starting point in the exploitation process and find that the vulnerability is actually fixed. This tends to remove a lot of the false positives produced by automated tools.
Sure, weeding out false positives does consume a fair amount of time in a penetration test. However, I wouldn't consider it a downside. It is just part of the whole penetration testing process. It probably will not cause any delays in the penetration testing process or affect it's quality as a good penetration tester should have already taken into account these factors at the start of the test.
A good penetration tester uses the tools available to him to make his job more efficient. Automated tools are no replacement for proper skills, knowledge and experience.